higgs phenomena in asymptotically free ......asymptotic freedom in gauges theories with scalars. as...
TRANSCRIPT
SLAC-m-1340 (T) November 1973
HIGGS PHENOMENA IN ASYMPTOTICALLY FREE GAUGE THEORIES*
T. P. Cheng
Department of Physics University of Missouri at St. Louis, St. Louis, Missouri 63121
and
E. Eichten and Ling-Fong Li
Stanford Linear Accelerator Center Stanford University, Stanford, Calif. 94305
ABSTRACT
We examine in detail the possibility of using the Higgs mechanism
to remove the catastrophic infrared singularities in non-Abe&n gauge
theories which are symptotically free. Our investigation encompasses
theories based on SU(N) or O(N) with scalars in one vector, two vector,
M vector, adjoint, tensor and adjoint plus one vector representations.
We find that for these theories an S-matrix, in the perturbative sense,
and asymptotic freedom can not coexist. We show that a wide class
of Yukawa couplings can be ignored in studying the large momentum
properties of the scalar couplings.
(Submitted to Phys. Rev.)
* Work supported by the U.S. Atomic Energy Commission.
I. INTRODUCTION
Recently Politzer , 1 Gross and Wilczek2 have discovered3 using the Gell-
Mann and Low renormalization group techniques, 4 that for non-Abelian gauge
theories the origin of the coupling constant space is a stable fixed point5 in the
deep Euclidean limit. Theories having this property are now often referred to in
the literature as being “asymptotically free”. 6 Furthermore extending the re-
sult of an earlier work of Zee, 7 Coleman and Gross8 have shown that no re-
normalizable field theory can be asymptotically free without non-Abelian gauge
fields. There are also theoretical arguments’ indicating that Bjorken scaling
as observed in the deep inelastic electroproduction experiments at SLAC and its
implied canonical behavior of the light cone in the configuration space can only
be obtained in a theory where the effective coupling constants vanish in this
asymptotic limit, All these developments lead us to the following conclusion:
if Bjorken scaling, as a true asymptotic phenomena, 10 is to be explained in re-
normalizable field theories, we must have a non-Abelian Yang-Mills theory for
the strong interactions.
The great practical advantages of asymptotically free gauge theories is
that one can study certain physically interesting strong interaction quantities
in the ultraviolet regime by perturbative methods. The high energy efe- annihi-
lation (via one photon) total cross section goes as s -1 with calculable logarithmic
corrections. 11 The anomalous dimensions of the entire tower of low twist oper-
ators in the Wilson light cone expansion may also be calculated. 12 It was found
that such theories give scaling up to logarithms. 13,14,15
-2-
One of the difficulties of asymptotically free gauge theories is the presence
of very severe infrared singularities coming from the massless nature of Yang-
Mills particles. Due to the non-Abel&n nature of these gauge symmetries, these
infrared singularities can not be handled like those of the usual QED. l6 The
consequence is that only off-shell Green’s functions can be studied and the
on-shell S-matrix elements are left undefined. However it is well-known that
the Higgs phenomenon, 17 in which the gauge symmetry breaks spontaneously, can
give masses to the gauge particles in such a way that the renormalizability of
the theory is preserved. The aim of this paper is to investigate whether by this
mechanism we can remove all the infrared singularities while maintaining the
asymptotic freedom.
The problem of incorporating scalars in the asymptotically free theories
was first examined by Gross and Wilczek. 13 The difficulty, as pointed out
by these authors, lies in the fact that the self-quartic couplings of scalars, in-
evitably present in any renormalizable field theories involving scalars, are in-
herently unstable but for the presence of gauge couplings. In Ref. 13, scalars
are restricted to a single low-dimensional representation. The cases of one
vector, one adjoint, or one symmetric tensor in SU(N) and the case of scalars
belonging to one (N, N) in SU(N)X SU(N) are worked out. Furthermore Yukawa
couplings are excluded D Since Yukawa couplings enter in a nontrivial way in the
renormalization group differential equations of scalar couplings (see Eq. (2.8)))
they can in principle have an effect on the stability properties of the scalar
couplings.
In section II we set up the general formalism and establish our notation.
We make some general remarks about the calculation of the coefficients in the
renormalization group equations for the coupling constants and discuss our
-3-
method of determining the stability of this system of differential equations., We
also breifly discuss the situation in semi-simple groups. We note that if any
one of the product groups is a U(1) the theory will not be asymptotically free.
In section III we consider the effect of the presence of Yukawa couplings on
the asymptotic freedom of the theory. We demonstrate that if the Yukawa cou-
plings are stable at zero they will be driven to zero at a faster rate than the
gauge coupling constant itself and hence will not influence the stability properties
of the scalar couplings. The proof for a case of more than one Yukawa cou-
pling is left to Appendix A.
In section IV we investigate the asymptotic stability of the scalar coupling
constants. We examine scalars belonging to one vector, two vector,
M vector, adjoint, antisymmetric tensor, symmetric tensor, and adjoint plus
one vector representations of O(N) and SU(N). Many of the details of the calcu-
lations are left to Appendix B. We observe a general pattern of the occurrence of
asymptotic freedom in gauges theories with scalars. As the number of scalar fields
increases, the dimension of the group and therefore the number of gauge bosons
must also be increased to maintain asymptotic freedom. We find no examples
when the symmetry is broken down to an Abelian symmetry which is also ultra-
violetly stable D For completeness we include an Appendix (C) that review the
results of one of us (L,F,L.)18 on the symmetry breaking induced by the Higgs
mechanism for various representations of SU(N) and O(N).
In section V we make some concluding remarks as to the compatibility of
asymptotic freedom and the existence of an S matrix in perturbation theory.
II. GENERAL FORMALISM
The most powerful technique used in studying the asymptotic properties of
the renormalizable field theory is the Callan-Symanzik equations. We refer the
-4-
reader to Coleman’s Erice lecture 19 for an elegant presentation of this technique.
Here we write down only those differential equations for the invariant coupling
constants which are needed for studying whether the gauge theories can be
asymptotically free in the presence of the Higgs phenomenon.
Let Ais pi, +a be the hermitian gauge fields, real scalar fields and
spin one-half fields respectively. The most general renormalizable Lagragian,
which is locally gauge invariant under some simple compact Lie group $2 with
structure constants C abc , is as follows:
where V(G) is some quartic polynomial in $
(2.1)
I V(@) = $ f.. 1jM ‘i’j’k@Q + lower order terms (2.2) .
and
F” = aAa-8A” -gC abc Ab A~ PV v I-1 I-J v P v (2.3)
(2.4)
(2.5)
The scalar (fermion) fields belong to some, in general reducible, representa-
tion of 9, with corresponding representation matrices of the generators,
Ba (ta ij alp ). The constraints on Oa, t”, m9, h and V due to requirements of
hemiticity and gauge invariance of the Lagrangian may be easily worked out.
Given the above Lagrangian we can immediately write down the lowest order
approximation of the renormalization group equation for the effective coupling
-5-
constants.
16x2 k = - dt + S1(G) - $ S3(F) - 6 S3( S) g3 E - ibOg I
dh. 16n2 --$ = 2 hmhihm + i ( hmhmhi + hihmhm) + 2 Tr (hihm) hm
- 3i2tahita+ S2(F)hi] g2 + A f.. f 288~r~ rjkQ jkQm hm
16~~ % zz f ijmnfmnkQ ’ fikmnfmnjl ’ fiQmnfmnjk
- l2 ‘2(s) g2 fijkQ + 3A ijke g4 + 8 Tr(hihm) fmjM
- 12H ijkQ
where
A Es ijkQ ba, ebJj lea, eblk, + (ea, eb& ba, eblje
+ (ea, eb& {ea, ebljk
and
H = ijkQ - & Tr (hi’j ihk, hjI + hihk (hj, hl\ + hihQ{hj, hk/ )
We have used the following identities:
C acd Cbcd = S1(G) 6ab
(tata)ij = S2(F) 6.. 1J
(eaeajij = s2(s) bij
tr(tatb) = S3(F) 6ab
tqeaebj = s3(s) 6ab
(2. ‘3
(2.7)
(2.8)
(2.9)
(2.10)
-6-
which define constants S1, S2 and S3 depending only on the group (G) and the
representation of the fermions (F) and scalars (S). We have also used the fact
that f ijke is totally symmetric.
A number of comments concerning this set of coupled equations are in order.
(A). These equations apply in the deep Euclidean region where all the
dimensional coupling constants, like mass term and super-renormalizable inter-
actions, can be dropped. They describe the response of the coupling constants
when the normalization point M at which the couplings are defined is changed to
hM. Thus the coupling constants appearing in the above equations are functions
of t = -Qn A and are commonly referred to as “effective coupling constants” or
“running coupling constants”. The boundary condition for these effective cou-
pling constants is that they are equal to the couplings appearing in the Lagrangian
defined at Euclidean symmetric points p2 = - M2. For simplicity of notation we
shall not give the effective coupling constant separate labels and their dependence
on t is to be understood in the following discussion.
(B). Since we are only interested in exploring around the origin of coupling
constant space, it is adequate for our purposes to calculate to lowest order.
It should be noted that the lowest order terms are not necessarily all single loop
diagrams. In particular, Fig. 2c is a two-loop term. However, as it turns out,
the structure of Eq. (2.8) is such that if the theory is to be asymptotically free
the scalar couplings f must be proportional to g2 as g approaches zero in the
ultraviolet limit. Consequently, the hff term may be dropped and only one-loop
diagrams need be considered.
(C). The algorithm for the perturbation calculation of the right hand side of
Eqs. (2.6-2.8) is by now well known. It involves calculating the logarithmically
divergent parts of diagrams which contribute to the coupling constant
-7-
renormalization; and then taking the logarithmic derivative with respect to mass
scale M, These lowest order equations are gauge independent. 13 The computa-
tions are simplest when done in Landau gauge where many one-loop &grams are finite and hence do not contribute. The relevant diagrams for equations (2.6),
(2.7) and (2.8) are displayed in Figs. 1, 2 and 3 respectively.
(D). We note that in the presence of spontaneous symmetry breaking via the
Higgs mechanism the perturbation calculations of the renormalization group
equations for g, hi and f.. 1.W
do not depend on whether or not the scalar fields
are shifted to have zero vacuum expectation values, (and thus giving masses
to the gauge bosons and fermions). This is because shifting the scalar fields
can only affect super-renormalizable interaction terms in the Lagrangian;
these terms do not affect the renormalization group equations.
(E) . As explained in Ref. 19 the renormalization group equation for a
single coupling constant, f, is of the form 2 = p(f) and has a stable%xed point
atf=OifP(f)jf=o=Oand glfZo<O. We first observe that Eq. (2.6)ca.n be
solved trivially to give
g2w = g2( 0) bO 1+ -
16~~ g2tf-x t
(2.11)
Hence if b. > 0, g(t) is real and tends to zero as t - I- to. Equation(2.7)will be
discussed in the next section; we will see that in the cases of interest that if
hi = 0 is a stable point as t - ~0 then hi = hi/g goes to zero as t - 00 . Therefore
the Yukawa terms will be ignored in the equation for the scalar couplings Eq.
(2.8). In order to find the stability properties of Eq. (2.8) we eliminate the g2(t)
16~~ dependence by introducing a variable u Z - bO
Q&f2P) + bO -t) and cjke - fijke/g2.
16r2 -8-
The stability equations for f ijke(u) may be written in the form
&i du = Fi(;;i (2.12)
where x’is a column vector with components the f.. l&Q
‘s and Fi(q are quadratic
functions of the components of z We find the stable fixed-points of this system
of differential equations by the following method. We first find the solutions of
the system of quadratic equations Fi($ = 0 vi. The real solutions of Fi($ = 08i
are called critical points of the differential equations, We now consider the
stability of each of these solutions. We define the slope matrix of Fi(G at To
by
‘Fi(X? 1 Dij(+ ax - - 1X’X (2.13)
j . 0
The point To is a stable fixed point if
(a) Fi(<) = 0 Vi
and
(b) All the eigenvalues of Dij(<) have negative real parts.
The simple way to understand these conditions is as follows. The local
stability of the critical point x0 - is determined by retaining in Fi(x) only the
linear term in F= 2 - To;
n
“‘i _ -- du c
j=l Dijt’Ol 5 j (2.14)
Using the standard method, we write
ti= Ciehu (2.15)
-9-
to get the algebraic equations
Xi= Dij(yo)Cj (2.16)
To have non-trivial solutions, we have to demand det(D(yo) - AI) = 0, which gives
n complex eigenvalues hI, h2” o . hn. It is then easy to see from Eq. (2.15) that
the critical point x0 is locally stable if all the Ai’s have negative real parts so
that ti - 0, or x -2 as u - 00, 20 0
We note that To being a stable fixed point implies that f.. W
y- g2(t) 5 goes
to zero as t-cm. In most of the cases we consider in section IV we use computer
to do these straightforward but tedious calculations.
(F). Up to this point we restricted our considerations to simple Lie groups,
1. e., theories with only one gauge coupling constant. We shall make a brief
comment on the more general cases involving semi-simple groups; which are
just direct products of simple groups G1 X G2 X . . . X Gn, each with its own coupling
constant gi. To lowest order (gf ) the renormalization group equations for each
of the coupling constants is independent of other coupling constants 21 and therefore
the results can be deduced from that of simple groups. It is interesting to note
that if one of these groups, Gi, is an Abelian (U( 1)) group, the associated gauge
coupling will not be driven to zero and the theory is not asymptotically free.
Since most unified theories of weak and electromagnetic interactions are based
on the gauge group SU(2) X U(l), the ultraviolet behavior of such theories is not
controlled by the fixed point at zero coupling and the Johnson-Baker-Willey and
Adler programs remain a possible approach to Q. E. D. within these unified
theories.22
- 10 -
III. RENORMALIZATION GROUP EQUATION FOR THE YUKAWA COUPLINGS
No renormalizable field theory can be asymptotically free with non-Abelian
gauge fields. It is understood in this statement that one accepts the semi-
classical arguments that the quartic self -coupling f.. w
must be a positive defi-
nite tensor for the energy spectrum to be bounded below. On the other hand,
Symanzik has suggested that such argument may not be so compelling in a full
quantum field theory 23 (iO e., beyond the tree graph approximation). Clearly,
if this positivity condition is relaxed, one can have an asymptotic free (p4 theory.
However, it should be pointed out that in the view of the works by Zee, 7 Coleman
and Gross8 it is impossible to construct any realistic theory even if the sign of
$ term can be changed. The nontrivial introduction of fermions will immediately
destroy the asymptotic freedom since Yukawa couplings are inherently ultra-
’ violet unstable, independent of the choice of sign for the (p4 term.
The asymptotic stability property of Yukawa couplings is altered dramatically
in a non-Abelian gauge theory. As we shall see, Yukawa couplings can now be
driven to zero in the deep Euclidean limit. In fact, they are driven to zero at
a much faster rate than the gauge coupling itself.
To illustrate our point, consider first the simple solution where only one
Yukawa coupling constant is present. The set of coupled equations in (2.7) is
reduced to (recall that the hf2 term is negligibly small)
16n2 $ = Ah3 - Bg2h (3.1)
where A and B are some positive constants determined by the representation
content of the fermions and scalars. For example, the case where scalars
belong to adjoint representation is just the generalization of Yukawa coupling
scheme of Ref. 7 to the case of gauge theories. 24 To solve this equation it is
- 11 -
convenient to use the variable 7; = (h/g)2 to rewrite it in the form
87r2 2 bO = I; [AT; -(B--+1
where u and b. are defined by Eqs. (2.12) and (2.6) respectively. bO
If B - 3;<0,
there is only one critical point K = 0, because h = (h/g)2 is a positive quantity.
This case is unstable because the slope at this point is positive. We have to
restrict to the case B- bO
b T> 0, where there are two critical points 1; = 0 and
h = (B--$)/A. The stable fixed point is at h= 0 because the slope at this point
is given by - (B bO
bO
-2) < 0. This shows that h must approach zero at a faster rate
thangifB-_2>0.
We now turn to cases where there are more than one Yukawa coupling. We
consider the situation in which all the fermions belong to the fundamental repre-
sentation of the group. Consequently, the only type of Yukawa couplings is the
“vector-vector-adjoint” type, but the coupling constants for fermions in differ-
ent fundamental representations need not be the same. For definiteness, let us
say there are M such sets. The scalars may belong to a reducible representa-
tion, but we stipulate that there is only one adjoint. Thus there are M(M-1)/2
Yukawa couplings. We label them as follows:
h = hapta (3.3)
where ta, as before, is the representation matrix of the generators. The index
a runs from 1 to the order of the group. The indices a! and ,0 denote the “fermion
type’ ’ , namely they distinguish the different sets of fermions. So we can view
h QP
as an MX M symmetric coupling matrix. Writing the Yukawa couplings in
- 12 -
I
this form, we have for Eq. (2.7)
167~~ 2 = A(hlit~)~~ + Tr(hh) hop - Bg2h G (394)
A and B are again positive constants. For example in SU(N), A = (N2 - 3)/2N
and B = 3(N2 - 1)/N. The stability properties of this set of coupled equations can
be determined after they are simplified by diagonalization of the h “P
matrix.
Again, we find (see Appendix A) that if the theory is asymptotically free, i.e. ,
h w
/g- constants, then these constants are all zeroes.
The equations for the effective scalar self-coupling constants, Eq. (2.8),
are of the form:
g = af2+bfg2+cg4+dfh2+eh4 (3.5)
Since h goes to zero faster than g, i.e, , h <g, in the large ultraviolet regime,
the fh2 and h4 terms clearly can be dropped when compared with fg2 and g4 terms.
- 13 -
IV. RENORMALIZATION GROUP EQUATIONS FOR THE SCALAR COUPLINGS
In this section, we investigate the large momentum behavior of the scalar
couplings for those classes of gauge theories based on the familiar O(N) and
SU(N) groups with various choices of the representation of the scalars up to
second rank tensors. Tensors higher than 2nd rank are more difficult to handle.
But we note that in Eq. (2.6) for the gauge coupling constant the contributions of
gauge particles ,5$(G) have the right sign for asymptotic freedom and is propor-
tional to N in both O(N) and SU(N), while the scalar contributions S3(S) have the
wrong sign and are proportional to (N) k-l for scalars belonging to k-th rank
tensors (see Appendix B Table). For large enough N, any tensor higher than
second rank will destabilize the gauge coupling and can be ruled out for our
purposes.
Since the calculations are very similar in all cases, we will only give the
results here and give some of the details in the Appendix B. We will first
discuss the situation in O(N) in IV(A) and briefly describe the similar situation
in SU(N) in IV(B), followed by a discussion in IV(C).
A. Stabilities of Scalar Couplings in O(N)-
We first discuss the
O(N). The most general
pling constant;
simplest case with only one vector representation in
O(N) invariant quartic coupling contains only one cou-
N
q = - $ C(4,da12 a=1
The renormalization group equation for h takes the form
dh -L- [(N+8)h2 dt = l6X2 - 3(N-l)hg2 + ;(N-l)g41
(4.1)
(4.2)
-14-
or in terms of new variable 5; = h/g2.
1 - CQ zz p(x) = 1 pIi- 167r2
{(N+8)h2 + [bO-3(N-1)l T;+ ; (N-l)1 (4.3)
where b. is defined by Eq. (2.6)
dt and b. > 0 (4.4)
of Eq. (3.3) is a second order polynomial of the form
condition for p(x) = 0 to have real roots is simply
Since the right-hand side
p(h) = Ax2 + Bh + C, the 9
A = B” - 4AC 2 0. Let us call these two roots hl, h2 with h2 > A1 and calculate
the slope at these points
m dh h=hl = A(+A2) <O
!%!a I dh J=h2 = A(h2-hl) > 0
Hence the smaller root A1 is a stable fixed point. However from the condition
that the classical potential corresponding to this interaction is bounded below
we have to require h > 0. This constraint demands B < 0 because both A and C
are positive. This implies that it is most favorable to have b. as small as
possible in order to have large /B 1 to satisfy A = B2-4AC 2 0. This can be
achieved by having as many fermions as possible without changing the sign of bo.
It turns out that in all the cases we consider the results do not change very much
if we use the smallest possible b. instead of b. = 0. So for the sake of simplicity,
we will assume from now on that b. = 0. Then the discriminant condition is
simply
3(N-1)(2N-11) > 0 (4.5)
-15-
I
The theory is stable for N L 6. On the other hand it has been shown that for one
vector representation in O(N), the symmetry is broken from O(N) to O(N-1)
(see Appendix C) . This means that only for N=2 does it break the symmetry
completely, and for O(6) the smallest group which has the asymptotic freedom,
there are still ten massless gauge particles, corresponding to the generators
of a non-Abelian O(5) group.
We next consider the case where we use two sets of scalars Ga and $,,
each belonging to the vector representation of O(N). The quartic scalar inter-
action contains four coupling constants
- Lzint = (+a+a)t$,+i,)
(4.6)
We have imposed a discrete symmetry Ga - - $a to eliminate terms of the form
(@, +,)t$, ib) for simp1icity* Notice that we have to deal with four coupling
constants as compared to only one in the previous case. The renormalization
group equations for these coupling constants are also very complicated,
dhl 1 2 2 -= - dt 16n2
(N+8)hl + NA3 + 4 h3h4 + 4h4 2 - 3(N-1) hlg2 f ; (N-l)g4 1 (4.7a)
dh2 1 2 2 2 2 3 4 -= - dt 167i2
(N+~)A~ i- Nh3 + 4h3A4 + 4h4 - 3(N-l)h2g + 4 (N-l) g 1 (4.7b)
dh3 1 -= - dt 16~~
(N-I-2)(hl+A2)h3 + 2(hl+A2)h4 + 2Ai + 2At - 3(N-1) A3 g2 -t- i g4 1 (4.7~)
dh4 1 -=- dt 16n2
+ 4h3h4 + (N+2)A4 2 - 3(N-l)h4g2 + ; (N-2) g4 1 (4 a 7d)
-16-
We can proceed by using the variable hi = hi/g2 and finding all the real roots of
these four simultaneous non-linear equations and then we can check the stability
by calculating the eigenvalues of the slope matrix evaluated at these roots. Be-
cause of the complexity of all these calculations, we do all these steps numeri-
cally on the computer. The result is that this class of theories will become
asymptotically free for N 2 7 D But the symmetry is broken from O(N) to O(N-2)
by these two sets of vector representations(see Appendix C). Hence for Nz 7,
the Higgs phenomenon fails to remove all the infrared singularities.
When we compare the two cases we have considered so far, it seems that
as we put more scalars into the theory to break the symmetries further down,
the threshold value of Nfor asymptotic freedom of the theory also increases.
Since the number of coupling constants increases rapidly with the number
of vector representations, it is very difficult to do the case with more than two
vector representations even on the computer. However we observe that Eqs.
(4.7a-4.7d) is invariant under hl -h2, which corresponds to $, - i, in the
Lagrangian. The numerical solutions for these simultaneous equations have the
property hl = h2. If we impose the extra symmetry that the Lagrangian is in-
variant under $a - $,, then it implies hl = A2 and reduces the number of
equations to three, In the case where we consider M vector representations,
p , q)(2) . . . (#p) , “1 “2 aM
we can impose the interchange symmetry among
these M sets of scalar field to reduce the number of independent coupling con-
stants to only three. The final results are summarized in Table 1. This table
also shows the pattern we mentioned before. In the limit N is very large, the
theory is stable for M 5 0.85 N, but the symmetry is broken completely for
ML N-l =N.
-17-
We now consider the more complicated second rank tensor representations
of O(N). We only give the results here, Details of the quartic couplings and
stability equations are given in Appendix B. For the second rank symmetric
tensor representation, the asymptotic freedom starts from N=14. For the anti-
symmetric second rank tensor representation, it is asymptotically free for
N L 8. If we add a vector representation to this antisymmetric tensor, the
threshold value of Nfor asymptotic freedom starts to move up to N=9. But for
all these cases, it can be shown that the Higgs mechanism fails to remove the
infrared singularities.
B. Stabilities of Scalar Couplings in SU(N)
The situation in SU(N) is very similar as that of O(N). We only describe
the pattern here and refer the reader to the Appendix B for the form of quartic
couplings and stability equations.
In the simplest case of only one vector representation of SU(N), the theory
is asymptotically free for N 2 3, and the symmetry breaking pattern is from
sum - SU(N-1) 0 Hence SU(3) gauge symmetry has asymptotic freedom and
also has the infrared singularities associated with unbroken SU(2) symmetry.
We next consider the case of two vector representation, which is enough to break
the SU(3) symmetry completely. It turns out that the starting value for asymptotic
freedom moves up to n = 4, and SU(4) is reduced only to SU(2) by two vector repre-
sentations. For the cases with more than two vector representations, we impose
the same interchange symmetries among these vector representations as the
case in O(N). The results are summarized in Table 2. This shows the same
pattern as we indicated in the case of O(N). In the limit of large N we can solve
the equations analytically to show that it is unstable for N-l vectors in SU(N).
We next examine the cases with second rank tensor representations. The choices
-18-
of second rank symmetric tensor and adjoint representations have been studied
13 by Gross and Wilzek. We include their results here for completeness and com-
parison. In the case of the symmetric tensor representation, the theory becomes
asymptotically free for N > 9. - If we use only one adjoint representation, the
asymptotic freedom will start from N=6. When we add one vector representation
to the adjoint representation, the threshold for asymptotic freedom move up to
N=7. We have also considered an antisymmetric tensor in SU(N) ; we find asymp-
totic freedom for N 2 5. Again for these values of N there is always a non-
Abelian symmetry left after the spontaneous symmetry breaking through the Higgs
mechanism.
C. Discussion
The general pattern for all these cases seems to be that when we put in
scalars to remove all the infrared singularities, the theory will lose the asymp-
totic freedom. Or if we insist on asymptotic freedom, the maximum sets of
scalars we can put into the theory will only break the symmetry down to some
non-Abelian symmetry. For example, in the most familiar SU(3) group, it is
asymptotically free if we use only one triplet of scalars, but there is a non-Abelian
SU(2) symmetry left unbroken, which leads to the uncontrollable infrared catastrophe.
If we use two triplets of scalars to break the SU(3) symmetry completely, the
theory will no longer be asymptotically free.
We have not exhausted all the possible choices of representations for the
scalars. Since there are so many instances exhibiting the same feature, we
conjecture this to be a very general property of the Higgs phenomenon in asym-
totically free gauge theories. The validity of this conjecture and the possible
physical mechanism responsible for this property is under further investigation,
-19-
The fact that asymptotic freedom is only possible in the non-Abelian
gauge theories and not in any of the other renormalizable theories, (e.g.,
Abelian gauge theories, Yukawa coupling and h44 theories) is not well under-
stood. It is a possibility that the asymptotic freedom of non-Abelian gauge theory
is due to the presence of the infrared singularities, since there is no infrared
catastrophe in any other theories. All the cases we consider in the last two
sections, indicate that whenever the infrared singularities are removed or re-
duced to Abelian symmetry, the asymptotic freedom is lost.
V. CONCLUSION
The systematic investigation of the effect of scalar couplings on a wide class
of non-Abelian gauge theories leads to the results that the Higgs phenomenon
fails to remove the infrared singularities in such a way that the asymptotic
freedom is preserved. Whenever enough scalars are introduced to break the
gauge symmetries completely, the theory loses its asymptotic freedom. This
seems to indicate an intimate connection between the infrared singularities and
the asymptotic freedom in the asymptotic freedom in the non-Abel&n gauge
theories. It deserves further investigations to have a better understanding of
this feature and hopefully progress in this direction will shed some light on the
origin of the asymptotic freedom.
One might hope that somehow the symmetry is broken dynamically to give
masses to the gauge particles instead of the simple Higgs mechanism. So far
only the plausibility of this idea has been demonstrated in the context of Abelian
gauge theory. 25 However there is some difficulty in applying those arguments
to the more interesting non-Abelian gauge theory. One of the crucial assump-
tions needed to demonstrate the possibility of having spontaneously broken
-2o-
symmetry solution is that fermion self-energy X(p) has the asymptotic behavior
( 1
2 Z(p)- + y(g ), i.e., Z(p) has anomalous dimension. However in the asymp-
P totically free theory Z(p) behaves like (log p2,” , i. e., it can never have anomalous
dimension. 26 Hence if this approach is workable at all, new techniques are
needed to implement this idea.
In the absence of any reasonable physical mechanism to break the gauge
symmetries to give masses to the gauge particles, it seems that we have to face
the severe infrared singularities in the non-Abelian gauge theories seriously.
One has to handle the infrared singularities one way or the other in order to have
well-defined S-matrix elements. It has been speculated by a number of people
that the local gauge symmetry may in fact remain to be exact and the strong
coupling nature of the theory in the infrared limit may just provide us with the
desired mechanism for quark-confinement. 27 Although this possibility is very
b attractive, it has not been well formulated in any kind of theoretical framework.
Undoubtedly any progress along these lines will lead to a tremendous enhance-
ment in the understanding of strong interaction physics.
ACKNOWLEDGMENTS
It is a pleasure to thank T. Appelquist for his interest. One of us (T.P.C.)
would like to thank Professor Sidney Drell for his warm hospitality at SLAC
while most of this work was performed.
-21-
APPENDIX A
In this appendix we shall demonstrate that the only possible asymptotic
stable fixed-point of the equation for H =holp “P g
(see Eq. 3.4)
167r2 % bO
= A(HHH)afi + Tr(HH) Ha, - (B- 2) Help (A. 1)
is Hwtw) = O” Following the procedure for finding asymptotic stable fixed points outlined
in section II(E), we must first find solutions to the coupled nonlinear equations:
WHWaP + Tr(HH) H
@ - B’H
w = 0
(A. 2)
bO where B’ = B- 2 . Then we must determine for each solution of (A.2) whether
or not all eigenvalues of the corresponding derivative matrix, Eq. (2. ll), have
negative real parts.
We shall solve Eq. (A.2) by first diagonalize the hermitian matrix H w
In terms of the diagonalized matrix H’ w
=6 H o!!p 01’ Eq. (A.2) takes the simple
form
M
HJAH; + c p=1
H; - B’) = 0 a=1 , . . . M (A. 3)
For any CY, we have the choice that Hcl! is either zero or it satisfies the equation
M
A Hi + c H2 _ B’ = 0 p=1 p
(A. 4)
which gives non-zero solution for Ha!. The most general solution is then given by
L
A H,” + c H; - B’ = 0 p=1
for o=l, . . . L (A. 5)
and Ha = o for cr=L+l.. . M (A. 6)
-22-
with L = 0 , . . . 111. From (A. 5) we deduce that all the non-zero Hi are equal and
given by
Hi = H*2= --& for or=l.. . L (A- 7)
The next step is to show that in order for the derivative matrix to have all
eigenvalues with only nonpositive real parts we must have L=O. Namely, Hi (to)
are zero for all a! - the result asserted in section III.
The derivative matrix is of the form
ID = aP
CD. .D
D C . .
. . . L 0
i J
u -B’ c 8
B’ .
.
o -1 5’ I 0
(A. 8)
where C = (3A + L + 2) H*2 - B’
D = 2H*2
We note the LXL square in the upper left hand corner in ID matrix may be
written as
(C-D) I+ D T (A. 9)
where lC is an Lx L square matrix with each element equal to unity, hence
satisfing the identity
T2= L7r
-23-
(A. 10)
From this we conclude that the eigenvalues of ‘IJ? are either zero or L. Since the
trace is invariant, ‘If must be of the form
L 0
lr= 0
i 1 0
- 0
when diagonalized, Now the requirement that all the eigenvalues of ID a@
be
negative demand
(3A + L)(H*)2 + 2L(H*)2
(3A + L) (H*)2 - B’ < 0
-B’ <O
B’ < 0
(L> 1)
(A. lla)
(A. lib)
(A. llc)
Since H*2 = & and/B’ must be > 0 condition (A. lla) can not be satisfied, So
the only possible solution is L=O or Hz = 0 for all (Y. Since Hz is nothing but a
set of linear combinations of h* a!8/g , we conclude h:p/g- 0 as t - 00 .
-24-
APPENDIX B
In this appendix we will try to illustrate the calculations of the renormaliza-
tion group equations for the scalar couplings described in section IV. Since the
calculations in all the cases are very similar, we outline the steps in one case
and state the results for all the other cases.
(1). One vector and one antisymmetric tensor in O(N).
The scalars which belong to the vector representation of O(N) are denoted
by x ., i=l, . 0 . 1
N and second rank tensor representation are denoted by @. . , 11
i, j=l ,... Nwith $.. = -@... 11 11
We can write down the most general quartic self-
coupling by contracting all the indices to make it invariant under O(N),
(B. 1)
We also need the couplings between gauge particles and the scalars,
I[ “Xi - g Ap&Xk 1 (B. 2)
Where A i-G
=-A I-lji are the gauge particles in O(N). Though the tensor
notations oij, A . . PI
are very convenient for constructing invariance under O(N),
it is very clumsy in working out the Feynman rules, because two indices are
required to label them. To overcome this, we first make a transformation to
go over to one index labelling by writing
$ij = B?.@ 1J 01 '
A Pij
= BQ! AQ! ij P (B. 3)
-25-
where B 01, cu=l,2 . . . ,N(N-1)/2, are the complete set of real antisymmetric NX N
matrices with normalization
Tr(BoBP) = - i ,mp
Then the interactions (B. 1) and (B. 2) can be rewritten as
- izint = Al h3 8 ($,$,)2 + h2Tr(BaBPB?/BG) @“!@‘$y~s + 8 tXiXi12
+ h4 h5 4 (XiXi)(@a@a) + 4 (XiXj)(9a~~)(B~BP)ij
and
- $ (apxi - gAEByjxj) @xi - gA@B$k)
with
(B. 4)
(B. 5)
(B. 6)
With the interactions given above, it is straightforward to work out the Feynman
rules to calculate various renormalization constants. The renormalization group
equations are given by
-26-
dhl = -$ /[F 4 1
--a- h; + 2(2N-1) Alh2 + 6 A; -+ Nh; -I- 2h4h5 - 6(N-2) g2Al + gg4 1
dh2 1 -=- dt 167r2 i
(2N-1) A; + 12 Alh2 + f A; - 6(N-2) g2A2 + T g4 j
dh3 -L-i - = 16~~ 1 dt
(N+8) A; -I- v A; + (N-l) h4h5 + + (N-l) A; - 3(N-1) h3g2 + $ (N-l) ,
dhq = --& I[? +2] hlh4+(2N-1)h2Xq+ ~hlhg+h2h6+(N+2)h3hq+h3h5+2h~ dt
dh5 1 -rz - dt 16~~
2 hlh5 + (N-l) h2h5 + 2 h3h5 + 4h4h5 + 2
The calculations of these coefficients in the above equations are straightforward
but tedious. For example, consider the diagrams given by Fig. 4, where both
external lines and internal lines belong to antisymmetric tensor $a! with vertices
corresponding to A2 a These diagrams contain the internal symmetry factors
sa = c Tr(BQBPBPBS) S
Tr(BYB”BPBS ) P<
S
Sb = 1 Tr(BaBYBPB’)s Tr(BPBGBPBS) P5
S (B. 9)
,K-- SC = &
P5 Tr(BoBGBPB’)s Tr(BBByBPBS)s
-27-
where Tr(ABCD)s denotes the totally symmetric combination of A, B, C and D.
We need to sum over the intermediate states p, t and project out the term
Tr(B”BPBYBG)s in order to calculate their contribution to dh2/dt. This can be
done by the observation that BQ! , o!=l, 2 , . . 0, N(N-1)/2 form the complete set of
real antisymmetric NX N matrices. We can use them to expand any arbitrary
real antisymmetric NX N matrix M in the form
Mij = c BTj acr a
with M = -M.. ij 11
The coefficient aQ! can be calculated from Eq. (B.4)
O! = a - 2(MijBG) = -2(TrMB”)
(B. 10)
(B. 11)
Substituting back into Eq. (B. lo), we get
.
M.. = - 2 yA (Byj Byk) Mke i, j=l ,0-o, N 11
(B. 12) a!
Taking into account the antisymmetric nature of M, we can work out the complete-
ness relation,
c Ba Ba! = cd ij & - t ttpje - “jk”ie) (B. 13)
With this relation, the summation in (B. 9) can be worked out to give the coefficient
of Ai in the equation for dh2/dt. It is clear from this example how the calculations
are carried out. The case of an antisymmetric representation of O(N) can be obtained
from this example by setting all the coupling constant containing scalars h3, h4,
h5 equal to zero. Now we give the results in all the other cases.
-28-
(2). Symmetric second rank tensor representation in O(N) 0
The couplings are
with qLj = 4.. 11
and T: i
qii = 0
The stability equations are given by
dhl dt = --& 1 [y +7] + 2t2N2;3N-61hlh2+ + A; -6N$+9g4\
2N2 + 9N - -- - ( 36) 2 N
h2 - 6NA2g2+ 3N 2 g 41 f
(3). m vector representations in O(N).
We label these vector representations by Fl, F2, 0 D . Frn+ The couplings
are given by
m 2 =-
g c -gApabb pa p) (a 4 (9 - gA
i=l
We have assumed that Smt is invariant under any reflection Fi - - Fi to make
the thing as simple as possible. If we further impose the symmetry that gint
is invariant under any interchange among Fl . 0. Fm, we can get the stability
-29-
equations in the form
dh = -&- aI7 167r2
(Ni-8) A2 + (m-l)Np2 + 4(m-1) p 77 + 4(m-1) r2 - 3(N-1) g2h + 3g4kN-1) 1
I
2(N+2) ph + 477 A f N(m-2) p2 c 2p2 + 2q2 + 4(m-2) n p - 3(N-1) g2p f 32 4
ti= 1
dt 16n2 4hq + 2(m-2)q2 + 4qp + (N+2)p2 - 3(N-1) g2n +
where hi+, i=l, . . D m, p.. = p, i , j=l. 0 .m, n 11
ij=)7, i , j=l.. .m.
(4). Vector representation in SU( N) 0
We denote the complex vector representation by qi, i=l, 2.. 0n, and its
conjugate by ii = (qi)*, i=l. 0. N. The gauge particles are denoted by Aii, i, j=l.. .n
with Aii = (Avfj)* and c Aii = 0. Just like the case in O(N), the SU(N) invariant i
quartic and gauge couplings can be written down by contracting all the indices,
,‘g =--
g 2’ (apz/? + igAP: $j) (@pi - igA:i +k)
The stability equation is given by
dh = 1 dt 8712 i
tN+4) h2 _ 3(N;-1) hg2 + WW2+2N-2) g4
4N2
(5). Two vector representations in SU(N).
We denote these two vector representations by $ i and z/i . The quartic and
gauge couplings are
22 g = - $ (apqi + igAp: z+!I~) (ap$ - igALi iiik) - $ (a$; + igAji $;)(aP$‘i- igAEiGk)
-3o-
where we have assumed that .LZ int has the symmetry e1 - - qi .,
The stability equations are given by
dhl- 1 \ 75-2 1 (N+4) A; + Nh; + 2A3h4 + A; + A; - 9 3(N-1) (N2+2N-2)
4N2 g4
i
dh 2- 1 A-- (N+4)A;+2h;+2h3h4+h;+h; - + h2g2 +
3(N-l)(N2+2N-2) 41 dt 87r2 4N2
g I
dh3 _ 1 --- dt 87r2
A3 + A4 (Al 1 + h2) + 2hi -I- A; + A; - N;-1) 3( h3g2 + 30 g4 i 4N2 )
dh4 1 ( -= - dt 87r2 1
h4(h1+A2) + 4 h3h4 + N A; + (N+2) A; - 3(N2$) A4g2 + * g4 I I
dh5 _ -- dt A5
c (hl+h2) + 4h3 + 2(N+l) A4
I - + A5 g2
(6). m vector representations in SU(N).
These m sets of vector representations are denoted by i(l), $(2) . . . $(m).
The quartic and gauge couplings are
and m c
a=1 ta)j) (,l$taji _ igApik$ta)S
where we have assumed the symmetry under any reflection Qi ta) - - ita) . Again i
if we further assume the interchange symmetries among these m vectors, the
-31-
stability equations become
a= 1 dt 87r2
(N+4) A2 f (m-l) C Nq 2 + p2 + (r2 1 + 2(m-1) 77 p -
t 3( N-1)( N2+2N-2)
4N2 g4i
I
2(N+l) hq + 2hp f ,02 + a2 + 277 2 + N(m-2) n 2 -I- 2(m-2) 7 p -
+ 3(N2+2) g4 ’
4N2 t
Sk= 1 i
2hp f Np2 + (N+2)cr2 + 477 p + (m-2)p2 - 3(N;-1) pg2 + 3($;4) g4
dt 87~’
dr dt=
+B(N+l)p - + g2 +(m-2) c 1 (7) 0 One vector and one adjoint representations in SU(N).
We denote the vector representation by xi, i=l.. .N and the adjoint repre-
sentation by +i , i, j = 1. D .N with $i = ($i)* and F $: = 0 0 The quartic and gauge
couplings are given by
i k ’ ‘5 (X iX ‘) (llik $j)
zg = _ & (a xi + igAi * 2 !J - igAk x pi k ) - i (ap $J; + igAik$F - igAkj$)
X (apqj - igApj q’+ igA@ j i Q i i +,)
-32-
The stability equations are
dhl 1 _ -x-&z
6(N2+3) 2 A2 + Nh; + 2h4h5 - 6NAlg + 41 N2
9g 1
dh2 -=- dt + 6Nh2g2 + y g4
,
(N+4) A3 + 2 4 2 ?-I h2 + (N2-1) h4h5 + h2 (N-1)(N2+2N-2) _ 3(N2-1) h g2
N 5 4N2 N 3
+ 3(N-1)(N2+2N-2) g4
4N2
dh4 -= dt h2h4 + (N+l) A3h4 f
(8). Antisymmetric tensor in SU(N).
We consider the antisymmetric tensor representation of SU(N) for complete-
ness. (The symmetric tensor representation was covered by Ref. 13. ) We
denote the antisymmetric tensor representation by iij, i, j=l, . O. ,n with eij = - QjiO
Also the conjugate fields $Fj are denoted GijO The quartic and gauge couplings
are given by
-_(% =$
int
($j qij)2 + ~ iij ~ jk ike ~Qi
sg = - t (ap z+bij c igAL iQj + igALQ$ie)(8’ qji - igAy’ GQi - igAfQejQ)
-33-
I
The stability equations are
dh dt=
1 + 6 2 i N(N-1) 2 A2 + - N-1 2
2 71~ + & 77 _ 6(N+l;W Ag2
+ 3N2+8) g4
,
&I.= 1
dt 8n2 I 4An + y T2 _ fW+;W-2) 77 g2 + 12(N2-4N-16) g4
N
(9). Contributions of various representations to S3(R).
Using the trick of introducing the transformation matrix, as described above,
we can calculate easily the quantity S3(R), which is the contribution of scalars to
p(g) of the gauge coupling constant. We tabulate the results and show then in
Table 3.
-34-
APPENDIX C
We shall illustrate, in a simple case, the pattern of symmetry breaking via
the Higgs mechanism. We refer the reader to Ref. 19 for the details. One of
the purposes of this exercise is to emphasize that the form which vacuum expecta-
tion values of the scalars can take in a given theory is dictated by the structure
of the Lagrangian and cannot be set to some arbitrary values - a point which is
often overlooked in the various attempts of constructing gauge models of weak
and electromagnetic interactions. Let us consider a gauge theory based on the
scalars belonging to the vector representation of O(N). The corresponding
classical potential V($) is given by
A>0
and the coupling between gauge particles and scalars are given by
(C* 1)
(C-2)
where A Nj
=-A ~ji’ L j=L . . . , n are the gauge fields in O(N). The vacuum ex-
pectation value of $I~, <$I~ > , are chosen to be those which gives the absolute
minimum for the potential V(4). To get this minimum we can calculate its first
derivatives,
3v=
a4i
i=l, 2, . . . , N
The solution corresponding to the spontaneously broken symmetry is given by
(C.4)
Notice that the requirement of the minimum only determines the length of the
vector $+. This is because the potential only depends on the length of the vector z
-35-
We can choose it to be of the form
(C. 5)
All the other solutions are equivalent to this one because they can be reached by
O(N) rotation. The gauge particles obtain their masses through the coupling
given in (C. 2),
-+g” APij<Oj> Ayk +k’
With the solution given in (C. 5)) it is clear that A
massive, while A Pin’
i=l ’ ‘. ”
N-l become
~ij’ L j=l, . . . . N-1,remain massless, corresponding to the
generators of 0( N-l). The symmetry is broken from O(N) to O(N-1). An easy
way to understand this result is the observation that the vacuum expectation
value <$I~> given in (C. 5) is invariant under all the rotations leaving the Naxis
unchanged, which is the subgroup O(N-1). With this picture in mind, it is very
easy to get the results in the cases where there are more than one vector repre-
sentations. Consider the case with two such vector representations, sayTl and
T20 The O(N) invariant potential can depend only on the length of each vector
and the angle between them, IF11 , IT21 and 13, * T,l . The solutions for the
minimum fix up the magnitudes for these three variables. We can choose first
vector with only first component non-zero, and the second vector with first two
components non-zero in order to satisfy these conditions. The symmetry is then
reduced from O(N) to O(N-2). We can generalize this argument to any number of
vector representations with the results that O(N) - O(N-m) for m set of vector
representations. In particular, it takes (N-l) set of vectors to break the O(N)
symmetry completely. Exactly the same argument can be applied to the SU(N)
group.
-36-
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
REFERENCES
H. D. Politzer, Phys. Rev. Letters 30, 1346 (1973).
D. J. Gross and F. Wilczek, Phys. Rev. Letters 3, 1343 (1973).
That the slope of the /3 function at origin in non-Abelian gauge theories is
negative was also noticed by G. t’Hooft but not published.
M. Gell-Mann and F. Low, Phys. Rev. 95, 1300 (1954). The renormaliza-
tion group was discovered by E. C. G. Stueckelberg and A. Petermann,
Helv. Phys. Acta. 2, 499 (1953); its role in the Gell-Mann-Low analysis
is discussed by N.N. Bogolubov and D.V. Shirkov, Introduction to the
Theory of Quantized Fields (Interscience, New York, 1959), Chap. 8.
The importance of stable fixed points of the renormalization group equa-
tions for strong interactions has been emphasized by K. Wilson, Phys.
Rev. D3, 1818 (1971).
Such theories are also referred to as “stagnant theories” (Ref. 7) because
of the hydrodynamical analogy of the renormalization group equation (see
Ref. 19). A stable fixed point corresponds to a stagnant point in such a
situation.
A. Zee, Phys. Rev. D 1, 3630 (1970).
S. Coleman and D. J. Gross, Phys. Rev. Letters 31, 851 (1973).
G. Parisi, Nucl, Phys. B 59, 641 (1973). C. G. Callan and D. J.
Gross, “Bjorken Scaling in Quantum Field Theory”, Princeton preprint
(1973).
There is of course the possibility that the scaling we are seeing is only an
intermediate energy phenomena. For such an alternate view see, e.g. ,
M. Chanowitz and S. Drell, Phys. Rev. Letters 30, 807 (1973).
-37-
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
T. Appelquist and H. Georgi, “e+e- Annihilation in Gauge Theories of
Strong Interactions”, Harvard preprint (1973). A. Zee, “Electron-Positron
Annihilation in Stagnant Field Theories”, Rockefeller Report No. COO-2232B-
28 (1973).
K. Symanzik, Corm-n. Math. Phys. 23, 49 (1971). N. Christ, B. Hasslacher
and A.H. Mueller, Phys. Rev, DC, 3543 (1972).
D. J. Gross and F. Wilczek, “Asymptotically Free Gauge Theories - I”,
National Accelerator Laboratory preprint NAL-PUB -73/4g THY (1973).
D. J. Gross and F. Wilczek, “Asymptotically Free Gauge Theories -II”,
Princeton preprint (1973).
H. Georgi and H.D. Politzer, “Electroproduction Scaling in an Asymptotically
Free Theory of Strong Interactions”, Harvard preprint (1973).
F. Bloch and A. Nordsieck, Phys. Rev. 52, 54 (1937),
P. Higgs, Phys. Letters 12, 132 (1964).
L.F. Li, “Group Theory of the Spontaneously Broken Gauge Symmetries,”
Stanford Linear Accelerator Center Report No. SLAC-PUB-1311 (1973) 0
S. Coleman, in the Proceedings of the 1971 International Summer School
“Ettore Majorana” (Academic, New York, to be published).
If some of the hi’s have vanishing real parts then one must go to higher
order terms in ki to decide the stability of the critic al point. This did
not occur for any of the cases we considered in this paper. For a dis-
cussion of this case and a general discussion of regional stability see for
example: N. Minorsky, Nonlinear Oscillations (Van Nostrand, Princeton,
N. J., 1962).
We emphasize that this result does not depend on the representation con-
tent of the Fermions or scalar fields. For example; if we had a set of
scalars transforming according to the (N, M) representation of O(N) x O(M)
-38-
then one might expect glgz and g2gs terms in the equations for pi(gi) from
graphs of type shown in Fig. lc 0 However simple calculation shows that
the coefficient vanishes due to the vanishing of the trace over internal
symmetry matrices.
22. See for example: K. Johnson and M. Baker, Phys. Rev. D g, 1110 (1973)
and the references contained therein. This remark clearly does not apply
to models where SU( 2) x U( 1) is embedded in a larger group with no U( 1)
factors,
23. K. Symanzik, Lettere al Nuovo Cimento 5, 77 (1973). However a refined
argument against h < 0 has been given by S, Coleman and E. Weinberg
Phys. Rev. D1, 1888 (1973), by using the renormalization group equation.
24. In the notation of Ref. 7, A = 2sl + s2 + 2s3, and B = 6sl + 3s2.
25. Y. Nambu and G. Jona-Lasino, Phys. Rev. 122, 345 (1961). H. Pagels,
Phys. Rev. D 1, 3689 (1973). R. Jackiw and K. Johnson, “Dynamical
Model of Sponteneously Broken Gauge Symmetries”, MIT preprint C. T. P.
No. 348. J. Cornwall and R. Norton, “Spontaneous Symmetry Breaking
Without Scalar Mesons”, UCLA preprint UCLA/73/TEP/78.
26. T. M. Yan (private communication). We would like to thank Professor Yan
for this remark.
27. See reference 13. Also S. Weinberg, Phys. Rev. Letters 31, 494 (1973).
- 39 -
1.
2.
3.
4.
FIGURE CAPTIONS
Lowest order (g3) contributions to the gauge coupling renormalization
constant defined here from the vector- fermion vertex. The solid directed
lines are fermions; the wavy lines are the gauge bosons; dashed lines are
scalars; and looped lines represent the Faddeev-Popov ghosts. Diagrams
la, lb, and lc correspond to the first, second and third terms on the
right hand side of Eq. (2.6). Other possible one-loop diagrams not dis-
played do not contribute in Landau gauge.
Lowest order contributions to the Yukawa coupling renormalization
constants. Diagrams in 2a, 2b and 2c correspond to the h3, hg2 and
hf2 terms in Eq. (2.7).
Lowest order contributions to the quartic self coupling renormalization
constants. Diagrams in 3a, 3b, 3c, 3d and 3e correspond to the f3, fg2,
g4, fh2 and h4 terms in Eq. (2.8).
The three graphs contributing to the Ag terms in the renormalization group
dh2 equations for dt , in the case that the scalar fields belong to the anti-
symmetric tensor plus one vector representation of O(N).
1.
2.
3.
TABLE CAPTIONS
The threshold values for asymptotic freedom in O(N) with various vector
representations and the pattern of symmetry breaking.
The threshold values for asymptotic freedom in SU(N) with various vector
representations and the pattern of symmetry breaking.
Values of S3(R) of various groups and representations.,
-4o-
TABLE 1
N Threshold
M” for Asymptotic
Freedom Symmetry Breaking
3 7 O(N) - O(N-3)
4 8 O(N) - O(N-4)
5 9 O(N) - O(N-5)
6 10 O(N) - O(N-6)
7 11 O(N) - O(N-7)
* Number of vector representations.
TABLE 2
M*
N Threshold for Asymptotic
Freedom Symmetry Breaking
3 5 SU(N) - SU(N-3)
4 6 WN) - SU(N-4)
5 7 SU(N) - SU(N-5)
6 8 SW-I - SU(N-6)
7 10 SU(N) - SU(N-7)
8 11 WN) - SU(N-8)
* Number of vector representations
TABLE 3
Representation S3(R) for O(N)
Vector l/2
S3( R) for SU(N)
l/2
Adjoint
Antisymmetric 2nd rank tensor
Symmetric 2nd rank tensor
N-2 2 N
N-2 2
N-b2 2
totally symmetric 3rd rank tensor
y (N+4) lNc2)(N+3)
4
Totally symmetric k-th rank tensor
JN+lHN+2) 0 l l W-k-2) (N+2k-2)
2(k-1) 1 jN+2)(N+3) l * l (N+k)
2(k-1) I
(a>
(b)
Fig. 1
1433Al
I/
- u
- 0 -
: !
/ ,“, .,
1
” /
I
.’
(a>
(b)
\ \
\ / \ \ \ r / y ’ / \ + 3 OTHER DIAGRAMS
/ \ / \ /’ \ \
\ \
hi)
(e>
\ \ / \ \ (3 ’ / k /\ + 3 OTHER DIAGRAMS
+ 5 OTHER DIAGRAMS
Fig. 3
/ ,Q, P / ‘8
( Id P
2433A 4
Fig. 4